APOE4 carriers and non-carriers with the clinical diagnosis of Alzheimer’s dementia and minimal amyloid plaques

Abstract

Importance

Amyloid-β (Aβ) plaques are a cardinal neuropathological feature of Alzheimer’s disease (AD), yet over a third of apolipoprotein E ε4 (APOE4) non-carriers with the clinical diagnosis of mild-to-moderate Alzheimer’s dementia may not meet positron emission tomography (PET) criteria for significant cerebral amyloidosis.

Objective

This study sought to clarify the percentage of APOE4 carriers and non-carriers with the primary clinical diagnosis of mild-to-moderate Alzheimer’s dementia near the end of life and minimal Aβ plaques at autopsy—and the extent to which these cases are associated with appreciable neurofibrillary degeneration or a primary neuropathologic diagnosis other than AD.

Design

Participants in this study were obtained from the National Alzheimer’s Coordinating Center’s Uniform Data Set (UDS).

Setting

The UDS comprises longitudinal clinical assessments performed at the Alzheimer's Disease Centers funded by the National Institute on Aging. Neuropathology data is available for the subset of expired participants.

Participants

Exactly 100 APOE4 non-carriers and 100 carriers had the primary clinical diagnosis of mild-to-moderate Alzheimer’s dementia at their last visit, known APOE4 genotype, died within the ensuing 24 months, and underwent neuropathologic evaluation.

Main Outcomes and Measures

Standardized histopathologic assessments of Alzheimer’s disease neuropathologic changes were the primary measures of interest in this study, specifically CERAD neuritic plaque density score, diffuse plaque density score, and Braak stage for neurofibrillary degeneration.

Results

37% of APOE4 non-carriers and 13% of carriers with the primary clinical diagnosis of mild-to-moderate Alzheimer’s dementia had nonexistent or sparse neuritic plaques. 44% of the carriers and non-carriers with minimal neuritic plaques had Braak stage III–VI ratings and 38% met neuropathological criteria for other dementia-related diseases.

Conclusions and relevance

More than a third of APOE4 non-carriers with the primary clinical diagnosis of mild-to-moderate Alzheimer’s dementia had minimal Alzheimer’s disease plaque accumulation in cerebral cortex and, thus may show limited or no benefit from an otherwise effective anti-Aβ treatment. Almost half of participants with a primary clinical diagnosis of mild-to-moderate Alzheimer’s dementia and minimal plaque accumulation had an extensive topographical distribution of neurofibrillary degeneration. Additional studies are needed to better understand and treat patients with this unexpectedly common clinico-neuropathological condition.

Introduction

Amyloid-β (Aβ) plaques are a cardinal neuropathological feature of Alzheimer’s disease¹ (AD) and clinical trials targeting this pathology are currently underway. Recent positron emission tomography (PET) studies have shown that more than a third of apolipoprotein E ε4 (APOE4) non-carriers with the clinical diagnosis of mild-to-moderate Alzheimer’s dementia may not meet criteria for significant cerebral amyloidosis,² while 20–30% of all those with the clinical diagnosis of AD may not meet cliniconeuropathological diagnostic criteria for AD.³ Further studies are needed to clarify the extent to which these findings are related to low PET tracer affinity for neuritic and/or diffuse Aβ plaques, or instead, truly reflect low levels of amyloid accumulation in cerebral cortex. If the latter, these studies could also help clarify the extent to which patients with a clinical diagnosis of mild-to-moderate AD dementia and minimal amyloid accumulation may have appreciable neurofibrillary degeneration or a primary neuropathological diagnosis other than AD.

The aims of this study were (i) to determine the percentage of APOE4 non-carriers with the primary clinical diagnosis of mild-to-moderate Alzheimer’s dementia near the end of life and minimal Alzheimer’s disease plaques as assessed by consensus histopathologic scoring (ii) to explore potential associations between Aβ plaques and appreciable neurofibrillary degeneration in both APOE4 non-carriers and carriers with the primary clinical diagnosis of mild-to-moderate Alzheimer’s dementia and minimal Aβ plaques, as well as their demographic and clinical features; and (iii) determine the frequency of a primary neuropathological diagnosis other than AD in APOE4 carriers and non-carriers with minimal Aβ plaques. While others have sought to confirm this finding in an autopsy cohort, previous studies have not looked at APOE4 carriers and non-carriers separately, an important feature necessary to emulate the imaging studies.⁴.

Methods

Study sample

The study sample comprised research participants from the 34 past and present National Institute on Aging (NIA)-sponsored AD Centers who were assessed with the Uniform Data Set (UDS) and had their data uploaded to the National Alzheimer’s Coordinating Center (NACC) between September 2005 and September 2012. The UDS contains clinical and demographic information on participants with cognitive impairment due to AD and other etiologies, as well as cognitively normal participants. Detailed descriptions of UDS data have been published previously⁵. UDS participants may also consent to autopsy, in which case neuropathologic features are assessed by consensus guidelines and recorded using a standardized neuropathologic assessment form, which is submitted to NACC. It is important to recognize that although NACC has made several revisions to its neuropathologic assessment form, the consensus guidelines for assessment of neuritic and diffuse plaques as well as for neurofibrillary degeneration were constant throughout the study period.

Study inclusion criteria were applied in the following manner: of the 2,288 UDS participants with an autopsy form, 1,834 died within 24 months of their last clinical assessment. Of these, 794 had a primary diagnosis of probable or possible AD dementia. Limiting the analytic sample to those with an MMSE score of 16 – 26 reduced the sample to 230 participants. Removing the 30 participants with missing APOE4 genotype information resulted in a study population of exactly 100 APOE4 carriers (11, 80, and 9 with the respective APOE 4/4, 3/4, and 2/4 genotypes) and 100 APOE4 non-carriers (86, 14 with the respective APOE 3/3 and 2/3 genotypes). See eFigure 1 for a more detailed sample size derivation.

Participants were further characterized in terms of their sex, education, age (at symptom onset, last clinical evaluation, and death), years from onset of cognitive symptoms to last evaluation, and years from onset of cognitive symptoms to death.

Clinical measures of interest included the MMSE and the Clinical Dementia Rating Sum of Boxes Score (CDR SOB).⁶ The CDR grades participants’ cognitive and functional abilities in six domains: memory, orientation, judgment and problem solving, community affairs, home and hobbies, and personal care. The clinician, incorporating input from the co-participant, evaluates impairment in each domain as none (0), questionable or very mild (0.5), mild (1), moderate (2), or severe (3). The scores for each domain are summed to create a Sum of Box score ranging from 0 to 18, with higher scores indicating more severe impairment.

Aβ plaque frequency was assessed using CERAD neuritic plaque score template.⁷ In the NACC Neuropathology Data Set Coding Guidebook, neuritic plaques are defined as plaques with argyrophilic, thioflavin-S-positive or tau-positive dystrophic neurites with or without dense amyloid cores. This 4-level rating was dichotomized into none or sparse (minimal plaques) vs. moderate or frequent. Diffuse Aβ plaque frequency also was graded using the CERAD plaque template. Neurofibrillary degeneration was assessed by Braak staging⁸.

As part of the NACC standardized form, the neuropathologist estimates the most likely primary cause of dementia. The frequencies of these etiologies were summarized for APOE4 carriers and non-carriers with minimal neuritic plaques and Braak stage 0–II, as well as Braak stage III–IV.

Neurochemical analyses

In post-hoc neurochemical analyses, we sought to further characterize soluble and fibrillar Aβ levels with greater sensitivity than is obtained using the consensus histopathologic methods, thus clarifying the extent to which they were truly free of increased concentrations of Aβ. Tissue samples from the 50 participants with no or sparse amyloid plaques were requested from the individual ADCs. Samples from 22 participants were provided by the ADCs. Parietal tissue was available for 19 APOE4 non-carriers and 3 carriers, and temporal tissue was available for 18 APOE4 non-carriers and 3 carriers.

Gray matter from the temporal and parietal lobes (200 mg) were each gently homogenized in 1600 µl of 20 mM Tris, 5 mM EDTA, pH 7.8 plus protease inhibitor cocktail (PIC; Roche, Manheim, Germany) using a Teflon tissue homogenizer (10 strokes). The homogenates were centrifuged at 45,000 rpm (250,000 × g) for 1 hour, 4°C in a Beckman TLA 50.4 Ti rotor (Brea, CA), the supernatant recovered and saved as the ‘soluble’ fraction. The Tris-insoluble pellet was re-homogenized in 1200 µl ml of 90% glass-distilled formic acid (GDFA) and centrifuged at 45,000 rpm (250,000×g) for 1 hr, 4°C in a Beckman TLA 50.4 Ti rotor. The supernatant was then dialyzed (1000 Da molecular weight cutoff) in deionized water 2 times (1 h, 4 L each) followed by 3 changes in 0.1 M ammonium bicarbonate (1 h, 4 L each) and flash frozen in dry ice/ethanol and lyophilized. The lyophilized proteins were reconstituted in 1000 µl 5 M guanidine-hydrochloride (GHCl), 50 mM Tris, pH 8.0 with PIC, shaken for 3 h at 4°C and centrifuged as described above, and the supernatant saved as the ‘insoluble’ fraction. Total protein in the soluble and insoluble fractions was determined with Pierce’s Micro BCA protein assay kit (Rockford, IL). Both Aβ40 and Aβ42 were measured with ELISA kits from Life Technologies Corp (Carlsbad, CA), following the manufacturer’s instructions: Aβ40 KHB3481 (minimal detectable dose <6 pg/ml), Aβ42 KHB3441 (minimal detectable dose <10 pg/ml). In addition, we also used the Aβ42 ultrasensitive KHB3544 kit (minimal detectable dose <1 pg/mL). For the purpose of this study and based on previous laboratory experience, total Aβ <1,000 pg/mg was considered low or negligible Aβ, total Aβ 1,000–100,000 pg/mg was considered a moderate level of Aβ, and total Aβ>100,000 pg/mg was considered severe Aβ consistent with AD.

Statistical analysis

First, characteristics of APOE4 carriers and non-carriers were summarized and compared. Next, stratifying by APOE4 carrier status, differences in clinical and neuropathologic features were explored for those with and without significant Aβ plaques. All statistical comparisons were made using non-parametric Wilcoxon rank tests for continuous measures and Chi squared tests for categorical characteristics; for comparisons involving small frequencies (<5 in at least one category), Fisher’s exact test was performed. As this is a descriptive study, an alpha level of 0.05 was applied, and all p-values are presented without adjustment for multiple comparisons. Analyses were performed using SAS 9.3.

Research using the NACC database was approved by the University of Washington Institutional Review Board.

Results

On average, both APOE4 carriers and non-carriers had approximately 15 years of education, and had MMSE and CDR SOB scores of approximately 20 and 8, respectively. They were also fairly evenly represented by men and women. Compared to APOE4 non-carriers, APOE4 carriers were slightly younger at onset of symptoms, last clinical evaluation, and death (see Table 1). There were no statistically significant differences for years from onset to last evaluation or death, demographic characteristics, or cognitive performance.

Table 1.

APOE4 non-carriers and carriers with the clinical diagnosis of mild-to-moderate Alzheimer’s dementia

	APOE4 non-carriers	APOE4 carriers	p-value
n	100	100
age at last clinical evaluation	85.2 ±9.6	83.3 ±7.4	.01
age at death	86.0 ±9.5	84.1 ±7.4	.01
age at onset of cognitive symptomsa	78.5 ±9.9	75.8 ±11.7	.05
years from onset of cognitive symptoms to last evaluationa	6.7 ±4.1	7.5 ±9.7	.94
years from onset of cognitive symptoms to deatha	7.4 ±4.0	8.3 ±9.7	.90
males / females	59/41	51/49	.26
education (years)	14.9 ±2.9	15.1 ±3.1	.60
MMSE at last clinical evaluation	20.3 ±3.2	20.6 ±3.2	.56
CDR SOB at last clinical evaluation	8.5 ±3.9	8.0 ±3.3	.48

^a3 e4 non-carriers and 1 carrier were missing data on age at onset of cognitive symptoms

NOTE: Ages, years, and clinical ratings are shown as mean ± SD

Of the 200 APOE4 carriers and non-carriers, 70 (35%) had a primary neuropathologic diagnosis other than AD. Of those 70, 11 had a primary diagnosis of AD present but insufficient for a diagnosis of AD and 7 had a diagnosis of normal brain. Thus, 26% met criteria for a neurodegenerative disease other than AD.

Among APOE4 non-carriers with the primary clinical diagnosis of mild-to-moderate Alzheimer’s dementia, 37% had minimal neuritic plaques. Similarly, 28% had both minimal neuritic plaques and minimal diffuse plaques, in combination. As shown in Table 2, APOE4 non-carriers with minimal neuritic plaques had lower Braak stages for neurofibrillary degeneration, on average, compared to those with moderate to frequent neuritic plaques (43% vs. 95% with Braak stage III – VI; p<.001). Of note, only 3 of the 16 with minimal neuritic plaques and Braak stage III–VI had Braak stage V–VI.

Table 2.

APOE4 non-carriers with the clinical diagnosis of mild-to-moderate Alzheimer’s dementia by post-mortem assessment of neuritic plaque density

	APOE4 non-carriers with no or sparse neuritic plaques	APOE4 non-carriers with moderate or frequent neuritic plaques	p-value
Neuritic plaque frequency	37 (27 no, 10 sparse)	63 (28 moderate, 35 frequent)
diffuse Aβ plaquesa			<.001d
no	19	0
sparse	9	4
moderate	2	14
frequent	4	42
Braak Scoresb			<.001e
0	3	0
I–II	17	3
III–IV	13	26
V–VI	3	34
age at last clinical evaluation	86.8 ±7.5	84.3 ±10.6	.42
age at death	87.7 ±7.4	85.0 ±10.4	.33
age at onset of cognitive symptomsc	80.6 ±8.4	77.3 ±10.6	.25
years from onset of cognitive symptoms to last evaluationc	6.1 ±4.2	7.0 ±4.0	.18
years from onset of cognitive symptoms to deathc	7.0 ±4.3	7.7 ±3.9	.33
males / females	20/17	39/24	.44
education (years)	15.1 ±2.4	14.7 ±3.1	.68
MMSE at last clinical evaluation	21.3 ±3.2	19.6 ±3.0	.02
CDR SOB at last clinical evaluation	7.2 ±3.2	9.3 ±4.0	.02

^a3 APOE4 non-carriers with no or sparse neuritic plaques were not assessed for diffuse plaques, and 3 APOE4 non-carriers with moderate to frequent neuritic plaques were not assessed for diffuse plaques

^b1 APOE4 non-carrier with no or sparse neuritic plaques was not assessed for Braak stage

^c1 APOE4 non-carrier with no or sparse neuritic plaques was missing information on age at onset of cognitive symptoms, and 2 APOE4 non-carriers with moderate to frequent neuritic plaques was missing information on age at onset of cognitive symptoms

^dFisher’s exact test for a difference in proportions of none or sparse vs. moderate or frequent diffuse plaques

^eFisher’s exact test for a difference in proportions of stage 0–II vs. III–VI

NOTE: Ages, years, and clinical ratings are shown as mean ± SD

Post-hoc neurochemical assays were performed for 19 of the 37 non-carriers with minimal plaques. In both the parietal and temporal regions, 2 non-carriers (11%) had moderate levels of combined ‘soluble’ and ‘insoluble’ amyloid Aβ. None had high levels of Aβ approximating those seen in neuropathologic AD. Thus, classification of participants using CERAD neuritic plaque score was similar to that ascertained neurochemically for Aβ peptides.

Compared to APOE4 non-carriers, a lower percentage of APOE4 carriers with the clinical diagnosis of mild-to-moderate Alzheimer’s dementia had minimal plaque scores: 13% had none or sparse neuritic plaques, and only 4% had both none to sparse neuritic and diffuse plaques. Neurofibrillary degeneration was also less extensive overall for APOE4 carriers with minimal neuritic plaques compared to carriers with moderate to frequent neuritic plaques (46% vs. 93% with Braak stage III – VI; p<.001). Full results of these comparisons are presented in Table 3.

Table 3.

APOE4 carriers with the clinical diagnosis of mild-to-moderate Alzheimer’s Dementia by post-mortem assessment of neuritic plaque density

	APOE4 carriers with no-sparse neuritic plaques	APOE4 carriers with moderate-frequent neuritic plaques	p-value
Neuritic plaque frequency	13 (2 no, 11 sparse)	87 (21 moderate, 66 frequent)
diffuse Aβ plaquesa			.001c
no	2	1
sparse	2	0
moderate	3	16
frequent	6	65
Braak Scores			<.001d
0	0	1
I–II	7	5
III–IV	5	27
V–VI	1	54
age at last clinical evaluation	87.6 ±6.5	82.6 ±7.3	.01
age at death	88.8 ±6.4	83.4 ±7.3	.01
age at onset of cognitive symptomsb	81.7 ±8.0	74.9 ±12.0	.02
years from onset of cognitive symptoms to last evaluationb	5.9 ±3.5	7.6 ±10.3	.58
years from onset of cognitive symptoms to deathb	7.2 ±3.6	8.5 ±10.3	.76
males / females	6/7	45/42	.71
education (years)	13.8 ±2.5	15.3 ±3.2	.04
MMSE at last clinical evaluation	20.0 ±2.9	20.7 ±3.2	.54
CDR SOB at last clinical evaluation	7.6 ±5.6	8.1 ±7.4	.73

^a5 APOE4 carriers with moderate or frequent neuritic plaques were not assessed for diffuse plaques

^b1 APOE4 carrier with moderate or frequent neuritic plaques was missing information on age at onset of cognitive symptoms

^cFisher’s exact test for a difference in proportions of none or sparse vs. moderate or frequent diffuse plaques

^dχ² test for a difference in proportions of stage 0–II vs. III–VI

NOTE: Ages, years, and clinical ratings are shown as mean ± SD

One of the 11 APOE4 homozygotes with the clinical diagnosis of mild-to-moderate AD dementia had only sparse neuritic plaques; this individual was characterized as having moderate diffuse plaques, Braak stage III–IV, and a primary neuropathological diagnosis of Lewy body dementia.

In addition, APOE4 carriers with minimal plaques were older than those with moderate to frequent plaques at onset of symptoms, last clinical evaluation, and at death (p=.02, .01, and .01, respectively). While this trend was also observed in APOE4 non-carriers, comparisons were not statistically significant.

In the post hoc analysis, 3 of the 13 APOE4 carriers were analyzed using tissue homogenates and immunochemical assays for ‘soluble’ and ‘insoluble’ Aβ. Of these, one participant was confirmed to have low tissue Aβ levels and two were had moderate Aβ levels. None had Aβ peptides at the level observed in participants with neuropathologic AD. Full results are provided in e-Tables 1 and 2.

Of the 20 APOE4 non-carriers with minimal neuritic plaques and Braak stage 0 – II, 15 participants received a primary neuropathologic diagnosis other than AD. Two had AD pathologic changes, but the features were considered insufficient to explain the clinical diagnosis of dementia. The remaining three participants had no distinct neuropathologic features that could have explained the dementia diagnosis and were determined to have "normal brain." See Table 4 for full list of primary neuropathologic diagnoses.

Table 4.

Primary neuropathologic diagnosis for no to sparse CERAD neuritic plaque density APOE e4 carriers and non-carriers

Primary NP diagnosis	Braak stage 0–II		Braak stage III–IV
	APOE4 non- carriers	APOE4 carriers	APOE4 non- carriers	APOE4 carriers
Normal brain	3a	0	1	1
AD	0	2	1	1
AD pathology present but insufficient for diagnosis	2	1	1	0
Lewy body disease	3	0	2	1
Vascular disease	5	0	2	1
FTLD	2	1	1	0
Hippocampal sclerosis	3	1	1	0
Rosenthal fiber encephalopathy	1	0	0	0
Nigral degeneration with focal tauopathy	1	0	0	0
Tauopathy NOS	0	1	0	0
Progressive supranuclear palsy	0	1	1	0
Senile Dementia with Tangles (Tangle only dementia)	1	1	0	1
FTD - NFT	0	0	1	0
Tauopathy/Diffuse grain disease	0	0	1	0

^aContains one subject with write-in response "Clinical dementia, no neuropathological substrate"

Similarly, of the seven APOE4 carriers with minimal neuritic plaques and Braak stage 0 – II, three were considered to have dementia resulting from AD pathologic changes, and the remaining four were thought to result from non-AD neuropathologic features.

Finally, substantial neurofibrillary degeneration (Braak stage III – VI) was observed in 43% of APOE4 non-carriers and 46% of APOE4 carriers with the clinical diagnosis of mild-to-moderate AD dementia and minimal Aβ plaques. Those with Braak stage III – VI were, on average, 2 years older at last clinical assessment than those with lower Braak stage. Yet, among those with higher Braak stage, those without Aβ plaques were older than those with moderate to frequent plaques (age at last clinical evaluation; p-value =.04 from Wilcoxon test).

In order to better understand whether the substantial neurofibrillary degeneration could be driving the cognitive impairment observed in those with mild amyloid, we performed a post hoc analysis of the persons in the same age group in NACC who were cognitively unimpaired (defined as a diagnosis of “normal cognition” and MMSE>26) at their last evaluation, died within the next 24 months, and had an assessment of plaque density and Braak staging. We found that 101 of the 155 cognitively unimpaired persons with known Braak stage had no more than sparse neuritic plaques. Interestingly, 35% (35) of the 101 cognitively unimpaired persons with no more than sparse plaques were in Braak stages III–VI. Although this is a smaller percentage than those with the clinical diagnosis of mild-to-moderate AD dementia and no more than sparse plaques (45%; Chi square test, p=.10), 35% indicates that this is still a frequent pathology in normals.

Discussion

37% of APOE4 non-carriers with a primary clinical diagnosis of mild-to-moderate Alzheimer’s dementia had minimal AD plaque scores. A much smaller percentage (13%) of APOE4 carriers had minimal AD plaque scores. Neurochemical assays for ‘soluble’ and ‘insoluble’ Aβ peptides were then applied to brain samples from 22 of the 50 brain donors with minimal AD plaque scores, confirming the low level of Aβ in most of these samples of cerebral cortex. These findings support the results of recent PET imaging studies suggesting that many participants who meet clinical criteria for mild-to-moderate AD dementia do not appear to have high levels of Aβ accumulation in cerebral cortex^2,9

Almost half (45%) of the participants with mild-to-moderate AD dementia (APOE4 non-carriers and carriers) and minimal AD plaque scores had topographically extensive cerebral neurofibrillary degeneration (Braak stage III to VI). This combination of AD pathologic features long has been reported in the literature under several names, including “tangle only dementia” or “tangle predominant dementia” to reflect discordance between the two cardinal histopathologic features of AD. Indeed, this combination of low levels of amyloid plaques yet moderate to extensive neurofibrillary degeneration was recognized in the recent NIA-AA guidelines for neuropathologic evaluation of AD; the panel of experts recommended that this constellation of features not be reported as AD, and that care be taken to exclude other tauopathies. Subsequently, a new diagnostic category for this combination of neuropathologic features, extended to include non-demented subjects with less-extensive neurofibrillary degeneration, has been proposed, Primary Age-Related Tauopathy (PART)¹⁰, and has engendered significant debate over whether this is distinct from AD or a variant manifestation of AD.^8,11,12 Although the data are limited, if we assume PET imaging showing insignificant cerebral amyloid and neuropathologic evaluation showing minimal AD plaque scores are equivalent, then some participants classified clinically as suspected non-amyloid pathology (SNAP) might possibly be tangle only dementia or PART in cases with less severe cognitive impairment.¹³

Inheritance of the APOE2 allele is associated with a lower risk of AD dementia,^14,15 suggesting that APOE2 may confer protection against AD¹⁶. Our sample of APOE2 carriers is too small to resolve this point; however, we did not observe a higher frequency of APOE2 in Braak III–VI (23%) vs. Braak 0–II (37%) in the presence of minimal plaques.

As is clear from our analysis, the vast majority of neuropathologic data from AD Centers derives from cases with advanced dementia. In a post-hoc analysis, we analyzed data from NACC participants with severe AD dementia (MMSE<16 or unable to participate in testing due to cognitive impairment). Among the 19 with mild amyloid, 6 (32%) had Braak III–VI. This percentage is similar to that observed in normals at NACC (35%), and is consistent with those previously reported by Dugger at al¹⁷ who found that 38% of individuals who were clinically normal at death met NIA-Reagan criteria for “intermediate” probability of AD with Braak stages III and above. Together, these findings lend support to the suggestion that a moderate to severe tauopathy may not in fact related to dementia, and may simply be a manifestation of healthy aging. Differences in the proportion of patients in the earlier and more advanced stages of dementia with no or sparse plaques could be related to subsequent plaque deposition in the advanced stages of AD, slower clinical progression (such that fewer individuals reached the advanced dementia stages in their lifetime), or other factors. Regardless, current clinical trials using either soluble or fibrillar Aβ-modifying treatments seem unlikely to benefit participants in this situation, even if they are shown to be effective in Aβ positive individuals. Development of new treatment strategies will be required to accommodate this pathologic heterogeneity within the clinical diagnosis of mild-to-moderate AD dementia.

Interestingly, APOE4 carriers with higher plaque densities were statistically significantly younger than APOE4 carriers with lower plaque densities at the time of death. The possibility that less Aβ burden is associated with a longer life expectancy in APOE4 carriers (or in non-carriers, in whom the differences were not statistically significant) would need to be confirmed in an independent cohort.

A previous analysis performed on the NACC database by Serrano-Pozo et al⁴ produced similar results for participants with mild-to-moderate AD who came to autopsy; they found that approximately 14% of these participants did not have amyloid pathology. We used an overlapping but larger subject pool, examined pathologic distributions within APOE4 strata, showed the particularly high prevalence of APOE4 non-carriers with a clinical diagnosis of mild-to-moderate AD dementia and minimal Aβ pathology, and performed a sub-study demonstrating the absence of appreciable fibrillar or soluble amyloid. Our findings in APOE4 non-carriers also are consistent with previously published florbetapir PET findings.^18,19

This study had several strengths, including the utilization of a well characterized sample of participants seen at multiple institutions, standardized clinical and neuropathologic forms, and a sensitivity analysis investigating whether those with minimal evidence of Aβ plaques might have had preferential increases in soluble or insoluble Aβ that had not been detected. There are also limitations to our study. First, there are potential differences in staining protocols across the ADCs. Some of the autopsies were performed using silver staining methods, thioflavin T, or Congo red, which may be less sensitive methods for detecting diffuse Aβ plaques than immunohistochemistry. Unfortunately, the staining method or methods used are not available in the NACC neuropathology database, so we were unable to account for this protocol variation. We did, however, re-examine tissue from multiple sites with standardized immunohistochemical methods for Aβ and phosphorylated tau and found considerably good agreement in identifying the presence of amyloid.

Second, there was some missing data. 162 MMSE scores were indicated as missing due to a cause other than cognitive impairment; however, it is possible that this group of participants did, in some way, affect the study sample. Only one participant out of 200 was missing data on Braak stage (.5%). There were, however, 30 participants (13%) who were excluded from our study due to missing APOE data. While it is possible that these participants could have different pathologic distributions than those included, they did meet all of the same clinical study inclusion criteria.

Finally, there is potential for selection bias in that all of these participants survived long enough to enter this study and also consented to autopsy. Thus, our study sample may not be representative of the general population, although they are likely similar to those older adults who would volunteer for observational research studies and/or clinical trials. Previous clinicopathologic studies using the NACC data have not found it necessary to adjust for selection bias due to decision to undergo autopsy.²⁰

In summary, we found that 25% of patients with clinical diagnosis of mild-to-moderate Alzheimer’s dementia, including 37% of APOE4 non-carriers and 13% of APOE4 carriers, have low cerebral cortical Aβ accumulation and that almost half of APOE4 non-carriers and carriers with minimal Aβ have extensive neurofibrillary degeneration. These findings suggest that a non-amyloidogenic variant resembling the clinical phenotype of Alzheimer’s disease may be more common than previously expected among research participants with mild-to-moderate AD dementia, particularly in APOE4 non-carriers, and they provide additional evidence to suggest that these individuals may not respond to treatments that target either fibrillar or soluble Aβ. Additional research is needed to further characterize the longitudinal natural history, clinical, biological, and genetic features of mild-to-moderate AD dementia patients with low Aβ and develop strategies other than Aβ-modifying agents to treat and prevent it.